[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Reporter API cleanup for 1.0

From: Greg Hudson <ghudson_at_MIT.EDU>
Date: 2003-12-28 22:44:08 CET

On Sun, 2003-12-28 at 16:04, Ben Collins-Sussman wrote:
> On Sun, 2003-12-28 at 13:04, Greg Hudson wrote:
>
> > I will also document the following bizarre and gross report behavior I
> > discovered just now: if you do an "svn update A/B" and there is a
> > deleted entry for A/B, the adm crawler will report:
> >
> > set_path("", rev-of-anchor)
> > delete_path("")
>
> Let me clarify something: the reporter vtable really ought to have a
> new function called "start_report()", to bookend finish_report().

Actually, there's a much simpler answer: report paths should be
anchor-relative. Since the anchor directory always exists and is never
switched, the top level of the report would always legitimately *be* a
set_path(), which would give the current reporter implementation a rev
to hang a transaction off of. Also, my new code could assume that edit
paths and report paths are always the same.

But, alas, no good way to get there from here.

> So really, that set of calls above isn't so hideous. In a better world,
> it would look like
>
> start_report(anchor-revnum)
> delete_path("") # delete the target of the update

Still somewhat hideous, because in most cases start_report() would
receive the target-revnum, not the anchor-revnum.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sun Dec 28 22:44:47 2003

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.